This dissertation will involve the conduct of two interdependent, but distinct, scientific investigations. The first investigation will be a conceptual analysis of methodological problems encountered in impact evaluations of maternal-infant programs. This analysis will be based on a quasi-experimental evaluation of the Maternity and Infant Care Project in North Carolina. The conceptual analysis will involve identification of key methodological problems; identification, selection, and application of the most rigorous and feasible evaluation methods; and conclusions regarding chosen techniques. The evaluation has been designed to determine the impact of the Project on the incidence of preterm and low birthweight deliveries, and fetal and infant mortality. It will be conducted within the frameworks of two quasi-experimental designs: A multiple time series and a post-test only comparison group design. Retrospective vital statistics data will be used in the evaluation. Weighted least squares techniques for categorical data, multiple regression analysis, and regression techniques for time series analysis will be used to test three major hypotheses. A major contribution of this study will be the generation of evidence regarding the effectiveness of Maternity and Infant Care projects, a core program in the current national effort to reduce infant mortality. Both investigations will contribute to the development and testing of a set of rigorous, economical, and administratively feasible methods for evaluation of maternal-infant programs. It is expected that these methods will expand administrative capabilities for evaluation, and encourage the routine conduct of valid assessments. The information generated by this dissertation is required for the planning of effective strategies to meet the Surgeon General's goal of a 35% reduction in infant mortality by 1990.